Merge pull request #3060 from googlefonts/unicode_collect

[subset] de-duplicate the logic that finds unicodes corresponding to …
This commit is contained in:
Behdad Esfahbod 2021-07-14 17:50:46 -07:00 committed by GitHub
commit 1060b04938
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
3 changed files with 54 additions and 76 deletions

View File

@ -1367,41 +1367,14 @@ struct cmap
for (const EncodingRecord& _ : encodingrec_iter)
{
unsigned format = (base+_.subtable).u.format;
if (!plan->glyphs_requested->is_empty ())
{
hb_set_t unicodes_set;
hb_map_t cp_glyphid_map;
(base+_.subtable).collect_mapping (&unicodes_set, &cp_glyphid_map);
if (format != 4 && format != 12 && format != 14) continue;
auto table_iter =
+ hb_zip (unicodes_set.iter(), unicodes_set.iter() | hb_map(cp_glyphid_map))
| hb_filter (plan->_glyphset, hb_second)
| hb_filter ([plan] (const hb_pair_t<hb_codepoint_t, hb_codepoint_t>& p)
{
return plan->unicodes->has (p.first) ||
plan->glyphs_requested->has (p.second);
})
| hb_map ([plan] (const hb_pair_t<hb_codepoint_t, hb_codepoint_t>& p_org)
{
return hb_pair_t<hb_codepoint_t, hb_codepoint_t> (p_org.first, plan->glyph_map->get(p_org.second));
})
;
hb_set_t unicodes_set;
(base+_.subtable).collect_unicodes (&unicodes_set);
if (format == 4) c->copy (_, table_iter, 4u, base, plan, &format4objidx);
else if (format == 12) c->copy (_, table_iter, 12u, base, plan, &format12objidx);
else if (format == 14) c->copy (_, table_iter, 14u, base, plan, &format14objidx);
}
/* when --gids option is not used, we iterate input unicodes instead of
* all codepoints in each subtable, which is more efficient */
else
{
hb_set_t unicodes_set;
(base+_.subtable).collect_unicodes (&unicodes_set);
if (format == 4) c->copy (_, + it | hb_filter (unicodes_set, hb_first), 4u, base, plan, &format4objidx);
else if (format == 12) c->copy (_, + it | hb_filter (unicodes_set, hb_first), 12u, base, plan, &format12objidx);
else if (format == 14) c->copy (_, it, 14u, base, plan, &format14objidx);
}
if (format == 4) c->copy (_, + it | hb_filter (unicodes_set, hb_first), 4u, base, plan, &format4objidx);
else if (format == 12) c->copy (_, + it | hb_filter (unicodes_set, hb_first), 12u, base, plan, &format12objidx);
else if (format == 14) c->copy (_, it, 14u, base, plan, &format14objidx);
}
c->check_assign(this->encodingRecord.len,

View File

@ -30,7 +30,6 @@
#include "hb-open-type.hh"
#include "hb-ot-os2-unicode-ranges.hh"
#include "hb-ot-cmap-table.hh"
#include "hb-set.hh"
@ -174,32 +173,14 @@ struct OS2
if (unlikely (!os2_prime)) return_trace (false);
if (!c->plan->prune_unicode_ranges) return_trace (true);
hb_set_t unicodes;
if (!c->plan->glyphs_requested->is_empty ())
{
hb_map_t unicode_glyphid_map;
OT::cmap::accelerator_t cmap;
cmap.init (c->plan->source);
cmap.collect_mapping (&unicodes, &unicode_glyphid_map);
cmap.fini ();
hb_set_set (&unicodes, c->plan->unicodes);
+ unicode_glyphid_map.iter ()
| hb_filter (c->plan->glyphs_requested, hb_second)
| hb_map (hb_first)
| hb_sink (unicodes)
;
}
/* when --gids option is not used, no need to do collect_mapping that is
* iterating all codepoints in each subtable, which is not efficient */
uint16_t min_cp, max_cp;
find_min_and_max_codepoint (unicodes.is_empty () ? c->plan->unicodes : &unicodes, &min_cp, &max_cp);
find_min_and_max_codepoint (c->plan->unicodes, &min_cp, &max_cp);
os2_prime->usFirstCharIndex = min_cp;
os2_prime->usLastCharIndex = max_cp;
_update_unicode_ranges (unicodes.is_empty () ? c->plan->unicodes : &unicodes, os2_prime->ulUnicodeRange);
_update_unicode_ranges (c->plan->unicodes, os2_prime->ulUnicodeRange);
return_trace (true);
}

View File

@ -231,31 +231,14 @@ _remove_invalid_gids (hb_set_t *glyphs,
}
static void
_populate_gids_to_retain (hb_subset_plan_t* plan,
const hb_set_t *unicodes,
const hb_set_t *input_glyphs_to_retain,
bool close_over_gsub,
bool close_over_gpos,
bool close_over_gdef)
_populate_unicodes_to_retain (const hb_set_t *unicodes,
const hb_set_t *glyphs,
hb_subset_plan_t *plan)
{
OT::cmap::accelerator_t cmap;
OT::glyf::accelerator_t glyf;
#ifndef HB_NO_SUBSET_CFF
OT::cff1::accelerator_t cff;
#endif
OT::COLR::accelerator_t colr;
cmap.init (plan->source);
glyf.init (plan->source);
#ifndef HB_NO_SUBSET_CFF
cff.init (plan->source);
#endif
colr.init (plan->source);
plan->_glyphset_gsub->add (0); // Not-def
hb_set_union (plan->_glyphset_gsub, input_glyphs_to_retain);
hb_codepoint_t cp = HB_SET_VALUE_INVALID;
while (unicodes->next (&cp))
for (hb_codepoint_t cp : *unicodes)
{
hb_codepoint_t gid;
if (!cmap.get_nominal_glyph (cp, &gid))
@ -268,6 +251,46 @@ _populate_gids_to_retain (hb_subset_plan_t* plan,
plan->_glyphset_gsub->add (gid);
}
if (glyphs->is_empty ())
{
cmap.fini ();
return;
}
hb_map_t unicode_glyphid_map;
cmap.collect_mapping (hb_set_get_empty (), &unicode_glyphid_map);
cmap.fini ();
for (hb_pair_t<hb_codepoint_t, hb_codepoint_t> cp_gid :
+ unicode_glyphid_map.iter () | hb_filter (glyphs, hb_second))
{
plan->unicodes->add (cp_gid.first);
plan->codepoint_to_glyph->set (cp_gid.first, cp_gid.second);
}
}
static void
_populate_gids_to_retain (hb_subset_plan_t* plan,
const hb_set_t *unicodes,
const hb_set_t *input_glyphs_to_retain,
bool close_over_gsub,
bool close_over_gpos,
bool close_over_gdef)
{
OT::glyf::accelerator_t glyf;
#ifndef HB_NO_SUBSET_CFF
OT::cff1::accelerator_t cff;
#endif
OT::COLR::accelerator_t colr;
glyf.init (plan->source);
#ifndef HB_NO_SUBSET_CFF
cff.init (plan->source);
#endif
colr.init (plan->source);
plan->_glyphset_gsub->add (0); // Not-def
hb_set_union (plan->_glyphset_gsub, input_glyphs_to_retain);
_cmap_closure (plan->source, plan->unicodes, plan->_glyphset_gsub);
#ifndef HB_NO_SUBSET_LAYOUT
@ -328,7 +351,6 @@ _populate_gids_to_retain (hb_subset_plan_t* plan,
cff.fini ();
#endif
glyf.fini ();
cmap.fini ();
}
static void
@ -439,6 +461,8 @@ hb_subset_plan_create (hb_face_t *face,
return plan;
}
_populate_unicodes_to_retain (input->unicodes, input->glyphs, plan);
_populate_gids_to_retain (plan,
input->unicodes,
input->glyphs,